(This publish is a part of my “This Week in AI” collection, which is common off-the-cuff market commentary. I’m not an investor in both MosaicML or Databricks)
$21M per worker. That’s the worth Databricks is paying for MosaicML — a complete of $1.3B for 62 workers (in Databricks inventory, and likewise consists of worker retention packages).
One factor is evident – in the event you’re going to be aggressively buying Generative AI startups, you’re going to must pay up
However it might grow to be low cost in the long run given the scale of the chance.
That’s as a result of, past any Generative AI capabilities, Databricks’ transfer must be understood within the broader context of its fierce rivalry with Snowflake.
(Some facets of that rivalry are somewhat foolish – they (mysteriously) scheduled their large conferences on the identical precise dates this week.)
Each are vying to be the brand new information clouds, essentially the most gigantic enterprise market.
It’s develop into apparent that profitable that market means being the very best information repository (warehouse, information lake, information lakehouse) AND the place the place enterprises do all their information science and ML/AI work.
Each Databricks and Snowflake have been beefing up their ML/AI capabilities, in 3 ways:
Homegrown: Databricks traditionally has had an ML/AI providing. It additionally just lately launched Dolly, its personal LLM.
Partnerships: for instance, Dataiku, the main pure play enterprise AI platform for World 2000 firms, is a prime ML/AI associate for each firms.
Acquisitions: Snowflake acquired Streamlit for $800M final 12 months and AI search firm Neeva for $150M this 12 months (excluding retention packages). Databricks has made a number of acquisitions as nicely, MosaicML being the largest one to date.
The chess (and thoughts) video games between Snowflake and Databricks are fascinating – and there’s much more coming! Within the meantime, large congrats to MosaicML and Databricks.
Observe: I’m simply utilizing the worth per worker as a proxy to point this can be a large acquisition for a nonetheless small firm. I don’t have direct information of what Mosaic had constructed thus far apart from what’s within the press, but it surely appeared promising. [Update: it was subsequently reported, after the time of writing, that Mosaic was at $20M ARR]